Assessing Driving Risk Using Internet of Vehicles Data: An Analysis Based on Generalized Linear Models.

Area of Science:

• Transportation technology
• Insurance analytics
• Data science

Background:

• Internet of Vehicles (IoV) advancements enable new Usage-Based Insurance (UBI) models.
• UBI products require accurate driving risk identification and driver classification.
• Existing methods may not fully leverage real-time IoV sensor data for risk assessment.

Purpose of the Study:

• To develop a method for calculating driving risk scores using short-term IoV data.
• To identify key driving behaviors indicative of risk.
• To classify drivers into different risk categories.

Main Methods:

• Utilized ordinary least square and binary logistic regression models.
• Analyzed short-term IoV sensor data, excluding accident and claims history.
• Focused on parameters such as driving speed, braking frequency, engine RPM, and accelerator pedal position.

Main Results:

• Established a positive correlation between driving speed, braking events, RPM, and accelerator pedal position with driving risk.
• Demonstrated the ability to differentiate between various classes of risk drivers.
• Validated the feasibility of using sensor data for immediate risk assessment.

Conclusions:

• Sensor data from IoV is crucial and feasible for driving risk analysis.
• The findings have significant implications for improving traffic safety.
• This approach can refine motor insurance underwriting and pricing strategies.